Oversized Broadcast Domain Causing Network Issues

Intermediate Troubleshoot

Question

A network engineer reports that a flat 10.0.0.0/16 network with 5,000 devices is experiencing high CPU utilization on all hosts, degraded throughput, and occasional network outages. Identify the root cause related to the broadcast domain size and recommend a remediation strategy.

Given

{
  "network": "10.0.0.0/16",
  "devices": 5000
}

Answer

{
  "problem": "Single /16 broadcast domain with 5,000 devices generates excessive broadcast traffic",
  "usable_hosts_in_16": 65534,
  "broadcast_packets_per_device_per_second_estimate": "1-5",
  "total_broadcast_pps_estimate": "5000-25000",
  "recommendation": "Segment into /24 subnets (254 hosts each) with Layer-3 routing between VLANs",
  "required_24s": 20
}

Try to solve it first, then click Reveal Answer

Step-by-Step Explanation

Step 1: Understand broadcast domain scope

In a flat Layer-2 network: - Every broadcast frame (ARP, DHCP, some routing protocols) is delivered to ALL hosts - 10.0.0.0/16 has 65,534 usable addresses - With 5,000 active devices, each ARP request reaches all 5,000 hosts

Step 2: Estimate the broadcast load

Typical broadcast sources per device per second: - ARP requests: ~0.5–2 pps (resolving new destinations) - DHCP: rare, but periodic renewals add up - NetBIOS/LLMNR (Windows): 0.5–1 pps - Spanning Tree BPDUs: 2 pps from switches - Total estimate: 1–5 pps per device

With 5,000 devices × 5 pps = 25,000 broadcast frames/second delivered to every NIC.

At 100 Mbps with 100-byte frames: - 25,000 × 100 bytes = 2,500,000 bytes/s = 20 Mbps consumed by broadcasts alone - This is 20% of a 100 Mbps link — significant overhead

Each host's CPU must process every broadcast frame (interrupt-driven). Modern hosts handle this better, but embedded/IoT devices may saturate their CPUs.

Step 3: Why it gets worse (broadcast storm risk)

If a switch loop forms (misconfigured uplink, failed STP), broadcast frames can circulate indefinitely:

Switch A → Switch B → Switch A → Switch B → ...

Each traversal duplicates the frame. Within seconds, the entire bandwidth is consumed by looping broadcasts — a broadcast storm. With 5,000 devices, a storm escalates extremely fast.

Step 4: Recommendation — VLAN segmentation

Divide 10.0.0.0/16 into /24 subnets, each as an isolated VLAN:

VLAN 10: 10.0.10.0/24  → Department A (254 usable)
VLAN 20: 10.0.20.0/24  → Department B
VLAN 30: 10.0.30.0/24  → Department C
... (20 VLANs cover 5,000 devices at ~250 per VLAN)

Add a Layer-3 switch or router for inter-VLAN routing.

Result: - Broadcast domain: 5,000 → ~250 devices per VLAN - Broadcast traffic per device: reduced by ~20× (5,000 → 250 peers) - Storm impact: limited to single VLAN, not entire network

Step 5: Verify after segmentation

# Before segmentation
tcpdump -i eth0 ether broadcast | pps_counter

# After segmentation — compare broadcast rate
# Should see <50 broadcasts/second in a well-sized VLAN

Recommended maximum broadcast domain size: 250–500 hosts per VLAN for optimal performance.

Related Terms

More Troubleshoot Worksheets